DOI QR코드

DOI QR Code

Ship Monitoring around the Ieodo Ocean Research Station Using FMCW Radar and AIS: November 23-30, 2013

  • Kim, Tae-Ho (Department of Remote Sensing, Underwater Survey Technology 21 Corp.) ;
  • Yang, Chan-Su (Marine Security and Safety Research Center, Korea Institute of Ocean Science and Technology)
  • Received : 2022.02.05
  • Accepted : 2022.02.18
  • Published : 2022.02.28

Abstract

The Ieodo Ocean Research Station (IORS) lies between the exclusive economic zone (EEZ) boundaries of Korea, Japan, and China. The geographical positioning of the IORS makes it ideal for monitoring ships in the area. In this study, we introduce ship monitoring results by Automatic Identification System (AIS) and the Broadband 3GTM radar, which has been developed for use in small ships using the Frequency Modulated Continuous Wave (FMCW) technique. AIS and FMCW radar data were collected at IORS from November 23th to 30th, 2013. The acquired FMCW radar data was converted to 2-D binary image format over pre-processing, including the internal and external noise filtering. The ship positions detected by FMCW radar images were passed into a tracking algorithm. We then compared the detection and tracking results from FMCW radar with AIS information and found that they were relatively well matched. Tracking performance is especially good when ships are across from each other. The results also show good monitoring capability for small fishing ships, even those not equipped with AIS or with a dysfunctional AIS.

Keywords

1. Introduction

The Ieodo Ocean Research Station (IORS) is a fixed ocean structure for meteorological and oceanographic observation constructed on a submerged reef, Ieodo, located 158 km away from Jeju Island (32.123°N, 125.182°E). Since the IORS is located close to the exclusive economic zone (EEZ) boundaries of Korea, Japan, and China, it is in an ideal location to monitor ships in the area, in the event of disputes between states over environmental issues or fishing activities. Illegal fishing activities in the EEZ cause serious damage to the marine environment and are a point of contention between the local economies of the countries involved. Furthermore, IORS is located in the area connecting East Asia to the Pacific Ocean. As such, it experiences heavy international vessel traffic. It is thus of crucial importance to monitor ships around the IORS in order to prevent illegal fishing and to protect the marine environment.

Remote sensing technology is mainly used for ship monitoring in the open sea and coastal waters. Various sensors like the space-borne Synthetic Aperture Radar (SAR), Automatic Identification System (AIS), and ground-based radar have been applied to maritime ship monitoring (Zhao et al., 2014). As ground-based AIS and radar are especially suited to ship monitoring because they can acquire continuous data. Hong and Yang (2014) used ground-based AIS to reveal that no flag vessels often appear in fleets around the IORS. The results also show that fishing and cargo vessels were the dominant ship type in these waters. In the use of AIS alone, however, the navigator must operate the equipment directly (Zhao et al., 2014); therefore, a method that fuses two sensors to monitor vessels was conceived. However, there are no experiments for ship monitoring using successive data around IORS due to its limiting operational conditions, such as power supply, accessibility, etc.

Ground-based radar can continuously monitor for a region of interest in real-time. High-frequency (HF) radars, which use the frequency band of 3.0-30.0 MHz, have been utilized in maritime surveillance studies, because of their wide coverage and continuous-time operation (Khan et al., 1994; Dzvonkovskaya et al., 2008; Maresca et al., 2014). Unfortunately, these radars have poor range and azimuth resolution as compared to microwave radars (also known as marine radars). The Frequency Modulated Continuous Wave (FMCW) technique is useful for distance measurement, and has good resolution from radar to target, as it calculates the distance from the difference between transmitted and received frequencies (Stove, 1992). Lower power is required to operate the FMCW radar system as compared to pulse radar (Kim et al., 2009; Stateczny and Lubczonek, 2015). The marine radar with FMCW technique is therefore extensively used to prevent ship collisions in areas of limited power supply, such as bridges, towers, and in smaller power boats (Stateczny and Lubczonek, 2015).

Recently, several algorithms for ship detection, identification, and tracking were developed using various sensors, including SAR, AIS, optical, radar, etc. (Won and Ouchi, 2011; Krol et al., 2011; Sudhir et al., 2012). Dzvonkovskaya et al. (2008) and Maresca et al. (2014) introduced an evaluation of ship detection and tracking algorithms based on the range-Doppler power spectrum using a low-power HF surface wave radar system, the WERA HF radar system, in conjunction with AIS data. Mecocci et al. (1995) developed an improved automatic target recognition system using real-aperture radar imaging. The algorithm of ship detection and tracking using FMCW radar imaging was based on a search window developed by Hong and Yang (2013). This algorithm detects 100% of large vessels, 85.38% of small vessels, and has a tracking rate of 100%. However, it was tested on a small number of vessels, in a close to a coastline. The ship represented in the radar image depends on its size and speed, the observation area, and the radar image resolution. In the case of ship tracking algorithms, it is necessary to consider the overlap of ships in radar images. Therefore, to successfully use FMCW radar for ship monitoring, the algorithm must first be improved through test operations over long durations, in the open sea, where there is heavy vessel traffic.

In this paper, we introduce ship monitoring results using FMCW radar image and AIS around IORS. FMCW radar image processing and tracking algorithms were used to generate ship positions from the image. In order to evaluate for ship monitoring accuracy, static information of AIS and detection results based on FMCW imaging were compared. This paper is organized as follows; Section 2 outlines the geographical characteristics of IORS. Section 3 represents the procedure for FMCW radar and AIS data processing. Section 4 discusses the ship detection and tracking algorithm, and the results are described in Section 5. The conclusions are presented in Section 6.

2. Ieodo Ocean Research Station

Ieodo is an underwater reef located at a depth of approximately 4.6 m, and 158 km away from Jeju Island (Fig. 1(a)). IORS was constructed at a depth of 40 m, 700 m to the south of Ieodo (Sim et al., 2004). Because the primary route ofships entering theYellow Sea and connecting EastAsia and the Pacific islocated close to Ieodo, it is geographically important area in maritime transport. Furthermore, analysisresults ofAIS data around IORS show that there are many small vesselsfrom various countries(Hong andYang, 2014). It is therefore very important to monitor ships around the IORS.

OGCSBN_2022_v38n1_45_f0013.png 이미지

Fig. 1. (a) The location of study area and (b) indicates coverage of FMCW radar (dot-line) and AIS (solid-line).

3. Data Acquisition and Processing

1) Configuration of Data Acquisition System

The data acquisition system used in this research comprises the FMCW radar, AIS, GPS, and heading sensor (Fig. 2). Analogue data from FMCW radar is converted into digital output by an interface box, and this data is saved in a computer system with heading sensor data through SimNet Network, which is network hub. AIS data encoded through an AIS receiver is stored in a computer system, and the position and direction information of FMCW radar are acquired using GPS and a heading sensor. In this research, we used Broadband 3GTM radar, a commercial radar intended for small vessels, manufactured by Simrad (https://www.simrad-yachting.com/en-GB/Products/ Radars/Simrad-3G-Broadband-Radar-en-gb.aspx).This radar uses the FMCW technique and the frequency X-band of 9.3 to 9.4 GHz. The FMCW radar has excellent target detection due to its high range resolution operates on very low power; about 18 W. The observation range can be set up from 0.05 to 44.45 km with 17 range options. The basic characteristics of the FMCW radar are given in Table 1.

OGCSBN_2022_v38n1_45_f0001.png 이미지

Fig. 2. Schematic of the data acquisition system on IORS.

Table 1. Characteristics of the FMCW radar (Broadband 3GTM radar)

OGCSBN_2022_v38n1_45_t0001.png 이미지

using GPS and a heading sensor. In this research, we used Broadband 3GTM radar, a commercial radar intended for small vessels, manufactured by Simrad (https://www.simrad-yachting.com/en-GB/Products/ Radars/Simrad-3G-Broadband-Radar-en-gb.aspx). This radar uses the FMCW technique and the frequency X-band of 9.3 to 9.4 GHz. The FMCW radar has excellent target detection due to its high range resolution operates on very low power; about 18 W. The observation range can be set up from 0.05 to 44.45 km with 17 range options. The basic characteristics of the FMCW radar are given in Table 1.

2) FMCW Radar Data

For the FMCW radar system used in this study, the reflected radar signals from scatters are saved as Spoke format to display image. The Spoke data is composed of header and retuned signals from targets in 4-bit format, and contains all the information required to reconstruct the positions of each target. The header data includes directions and covered range of each Spoke and pixel distance in the radar image. Examples of Spoke data structures are described in Appendix A.

3) AIS Data Processing

AIS data is saved in an encrypted ASCII format. These data have to be decoded in binary format to obtain the ships’ message ID, Maritime Mobile Service Identity (MMSI) and static and dynamic information. Message ID indicates the vessel’s type. For example, the ‘1’, ‘2’ and ‘3’ of message ID represent the dynamic information of class A type such as longitude, latitude, speed, heading and course of ground (COG), while ‘5’ of message ID includes the static information of class A type such as name, size, type, etc. MMSI data consists of 9 numbers, and the first three numbers mean unique country codes(Hong and Yang, 2014; Hong et al., 2018). In this study, we extracted information received by the AIS and classified the ship according to its nationality and type. Among the vessels equipped with class B type AIS, the ships that did not provide nationality information were classified separately.

4. Ship Detection and Tracking Algorithms

1) FMCW Radar Image Processing and Ship Detection Algorithm

FMCW radar signals include noise from its own transmitter or receiver and external interferences (Paprocki, 2011). Therefore, in order to optimize the results, noise-cancellation steps must be taken. The system noises appeared on each frame as the spokes containing strong radar signals in all cells (Fig. 3(a)). These were filtered by the threshold method using the number of cells which include maximum intensity (Fig. 3(b)). Moreover, consistent signals from nearby obstacles were masked (Fig. 3(c)). If only one frame consisting of 4096 spokes is used to distinguish targets from clutter, many false targets tend to be identified as ships. To prevent this, we generated the FMCW radar image at 1-minute intervals by combining the frames every 2 minutes. The images still contained clutter that was comparable to speckle noise form (Fig. 3(d)). The speckle noises were removed by the accumulation filter method, which uses the accumulated counting numbers in each pixel while combining the frames (Fig. 3(e)). Pixels retained after accumulation filtering were classified using the adjacency of each pixel (Fig. 3(f)) and size filtering was performed based on the area of each object (Fig. 3(g)).

OGCSBN_2022_v38n1_45_f0002.png 이미지

Fig. 3. Flowchart of image processing and detection algorithm.

2) Performance Comparison of Ship Tracking Algorithms Using Simulated Radar Images

In this study, simulated radar images were generated for three scenarios to define the main algorithm for ship tracking: (1) two vessels sailing close to a lighthouse, (2) two vessels passing by each other, (3) temporary target loss due to radar noise (Fig. 4). We adopted three methods, namely, the Block Matching Algorithm, Particle Filter, and Kalman Filter- that are widely used for tracking targets, and evaluated multiple objects tracking performance for the three cases. Our results showed that the Kalman Filter method has a good performance in tracking and computation time (Table 2). Hence, we have developed a Kalman Filter based tracking algorithm.

Table 2. Success orfailure withdifferent trackingalgorithms

OGCSBN_2022_v38n1_45_t0002.png 이미지

OGCSBN_2022_v38n1_45_f0003.png 이미지

Fig. 4. Performance comparison of ship tracking algorithm using three tracking algorithms.

3) Ship Tracking Algorithm

We developed a ship tracking algorithm to track the same object in successive radar images (Fig. 5). The algorithm includes a three-stage grouping process. Each grouping process compares the positions of the new (at Tn) and old (at Tn-1) ships. The ship detection results contain the position information of all the pixels constituting each object. If two ships include pixels of the same location, they are included in the one group. More than three ships included in the group indicates that the new ship overlapped with at least two previous ships. The number 1 and 2 (# = 1 or 2 in Fig. 5) illustrate cases where no ships overlap and only one exists, respectively. The first step of grouping is to extract the ships excluded from the tracking algorithm by comparing the new and previous ships. The second step compares new and previous ships. Note that the new ships inputted in the second step only include the ships in the groups which contain one or two ships. In the third step, the predicted ships are compared with objects in groups which has number 1 from the second grouping step. The predicted position of the ship is calculated using Kalman filter theory with position information for the past times.

OGCSBN_2022_v38n1_45_f0004.png 이미지

Fig. 5. Flowchart of ship detection and tracking algorithm using FMCW radar images.

OGCSBN_2022_v38n1_45_f0005.png 이미지

Fig. 6. Assessment of algorithm performance for overlapped ship tracks.

Fig. 6 shows the process of tracking the ships with overlapped tracks, using the developed tracking algorithm. If two ships are overlapped on the radar image, this object is excluded from the tracking algorithm (Fig. 6(c)~(d)). Then, when the ships are separated on the radar image, ship tracking is performed using predicted ship information using the Kalman filter (Fig. 6(e)).

5. Results and Discussion

We analyzed FMCW radar data acquired from 21th to 30th November, 2013, except for the observed data in the initial setup and test procedure. The FMCW radar provided continuous ship locations at one-minute intervals, which were classified by a multi-tracking algorithm. The AIS data also provided continuous position information of the vessel and were classified by identification number. The numbers of classified vessels are shown in Fig. 7. The number of vessels observed during the period were 930 and 990 for FMCW radar and AIS, respectively. During the observation period, the observed vessels on 23rd and 30th November were higher than other days. It is believed that the traffic decreased due to strong winds of more than 10 m/s during 24th to 29th November (Fig. 8). In particular, on 23rd November, the number of vessels observed through FMCW radar was relatively high. This is because ships without AIS information around IORS were observed through FMCW radar.

OGCSBN_2022_v38n1_45_f0006.png 이미지

Fig. 7. The number of observed vessels for FMCW radar and AIS during the 21th to 30th November, 2013.

OGCSBN_2022_v38n1_45_f0007.png 이미지

Fig. 8. The wind speed on the IORS during the 21th to 30th November, 2013.

In order to quantitatively evaluate ship detection and tracking results, a matching algorithm was applied using the trajectories of each of the separated objects. The matching results between radar and AIS showed an accuracy of approximately 22% (Fig. 9, Table 3). The matching ratios of cargos and tankers were relatively high, and the fishing boat was the lowest. This is because the extraction of ship data from FMCW radar is limited over long-distances, because insufficient scattering radar signals are reflected by the ship.

OGCSBN_2022_v38n1_45_f0008.png 이미지

Fig. 9. The number of matched and unmatched ship for each type throughdetection andtrackingalgorithms.

Table 3. Matching rate between ships from FMCW radar and AIS

OGCSBN_2022_v38n1_45_t0003.png 이미지

The detection performance of the radar is affected by the distance between the sensor and the target. Fig. 10 shows the number of ship detections along the IORS. In the region, within 20 km, the number of vessels detected by FMCW radar is larger than that of AIS, and in the region over 30 km, the number of AIS vessels is larger than that of FMCW radar. This is because the number of vessels detected by FMCW radar in the region between 0 ~ 20 km is larger than the number of AIS vessels, presumably because there are many vessels operating or not equipped with AIS, in this area.

OGCSBN_2022_v38n1_45_f0009.png 이미지

Fig. 10. The number of FMCW radar, AIS and matched ship for each distance from IORS.

The trajectories from FMCW radar and AIS around IORS are illustrated in Fig. 11. The track information of the ships extracted by applying the detection and tracking algorithms were relatively consistent with AIS data. In particular, ships located close to IORS and large vessels such as cargos and tankers, showed good agreement. We achieved stable tracking even when the track of a ship is crossed (Fig. 8(d)).

OGCSBN_2022_v38n1_45_f0010.png 이미지

Fig. 11. The ship trajectories from FMCW radar and AIS around IORS.

Fig. 12 shows the monitoring capability of FMCW radar and AIS for small fishing ships, even those not equipped with AIS or with a dysfunctional AIS. In Fig. 5(a), AIS only displayed information for five ships, while FMCW radar detected more vessels. This can also be seen on photos taken at the IORS (Fig. 12(b), (c)).

OGCSBN_2022_v38n1_45_f0011.png 이미지

Fig. 12. Monitoring result of FMCW radar and AIS for small fishing ships.

Although detection performance of FMCW radar and AIS monitoring system was limited, it was able to monitor small vessels within a 20-km-radius of IORS. In addition, relatively large vessels such as cargos and tankers showed high detection rates, and the matching ratio of radar and AIS was more than 45%. However, noise was sometimes mistakenly identified as a vessel depending on weathers conditions. Hence, further research is needed to prevent false alarms and improve the ship detection of FMCW radar.

6. Conclusions

In this study, we monitored ships around IORS using FMCW radar imaging and AIS in the period between 23th to 30th November, 2013. The radar signal is converted to an image format through pre-processing, using an improved ship detection and tracking algorithm, and implementing the ship’s detection and tracking. Although we were unable to extract the data of all the ships around IORS, comparisons between FMCW radar and AIS information show a good agreement. In particular, the results showed excellent tracking capability even if the tracks of two ships overlapped. Ships not equipped with AIS, or with dysfunctional AIS, were also detected. However, ships at great distances cannot be detected, and clutter is often misidentified as ships in bad weather conditions. To improve ship monitoring algorithms, more observational data from various sensors, such as SAR and the other radars, are required to improve the noise filtering technique.

References

  1. Corda, S. and G. Bell, 2009.TechnicalPaper:NavRadar NRP Class Description, Navico, Auckland, New Zealand.
  2. Dzvonkovskaya, A., K.W. Gurgel, H. Rohling, and T. Schlick, 2008. Low Power High Frequency Surface Wave Radar Application for Ship Detection and Tracking, Proc. of 2008 IEEE International Conference on Radar, Adelaide, SA, Australia, Sep. 2-5, pp. 654-659.
  3. Hong, D.B. and C.S.Yang, 2013. Algorithm Implementation for Detection and Tacking of Ships Using FMCW Radar, Journal of the Korean Society for Marine Environment and Energy, 16(1): 1-8. https://doi.org/10.7846/JKOSMEE.2013.16.1.1
  4. Hong, D.B. and C.S. Yang, 2014. Classification of Passing Vessels Around the Ieodo Ocean Research Station Using Automatic Identification System (AIS): November 21-30, 2013, Journal of the Korean Society for Marine Environment and Energy, 17(4): 1-9. https://doi.org/10.7846/JKOSMEE.2014.17.1.1
  5. Hong, D.B., C.S. Yang, and T.H. Kim, 2018.Investigation of Passing Ships in Inaccessible Areas Using Satellite-based Automatic Identification System (S-AIS) Data, Korean Journal of Remote Sensing, 34(4): 579-590. https://doi.org/10.7780/KJRS.2018.34.4.1
  6. Jeong, J. and C. Yang, 2016.Automatic Image Contrast Enhancement for Small Ship Detection and Inspection Using RADARSAT-2 Synthetic Aperture Radar Data, Terrestrial Atmospheric and Oceanic Sciences, 27(4): 463-472. https://doi.org/10.3319/tao.2016.01.01.01(isrs)
  7. Khan, R., B. Gamberg, D. Power, J. Walsh, B. Dawe, W. Pearson, and D. Millan, 1994.Target Detection and Tracking with a High Frequency Ground Wave Radar, IEEE Journal of Oceanic Engineering, 19(4): 540-548. https://doi.org/10.1109/48.338390
  8. Kim, J.Y., K.T. Chong, and T.Y. Kim, 2009. X-Band FMCW RADAR Signal Processing for Small Ship, Journal of the Korea Academia-industrial Cooperation Society, 10(11): 3121-3129 (in Korean with English abstract). https://doi.org/10.5762/KAIS.2009.10.11.3121
  9. Krol, A., T. Stupak, R. Wawruch, M. Kwiatkowski, P. Paprocki, and J. Popik, 2011. Fusion of Data Received from AIS and FMCW and Pulse Radar -Results of Performance Tests Conducted Using Hydrographical Vessels "Tukana" and "Zodiak", TransNav, the International Journal on Marine Navigation and Safety of Sea Transportation, 5(4): 463-469.
  10. Maresca, S., P. Braca, J. Horstmann, and R. Grasso, 2014. Maritime Surveillance Using Multiple High-Frequency Surface-Wave Radars, IEEE Transactions on Geoscience and Remote Sensing, 52(8): 5056-5071. https://doi.org/10.1109/TGRS.2013.2286741
  11. Mecocci, A., G. Benelli, A. Garzelli, and S. Bottalico, 1995. Radar Image Processing for Ship-Traffic Control, Image and Vision Computing, 13(2): 119-128. https://doi.org/10.1016/0262-8856(95)93153-J
  12. Paprocki, P., 2011. Impact of Internal and External Interferences on the Performance of a FMCW Radar, TransNav, the International Journal on Marine Navigation and Safety of Sea Transportation, 5(3): 325-328.
  13. Sim, J.S., I.S. Chun, and I.K. Min, 2004. Construction of Ieodo Ocean Research Station and its Operation, Proc. of the Fourteenth International Offshore and Polar Engineering Conference, Toulon, France, May 23-28, pp. 140-157.
  14. Stateczny, A. and J. Lubczonek, 2015. FMCW Radar Implementation in River Information Services In Pland, Proc. of 2015 16th International Radar Symposium, Dresden, Germany, Jun. 24-26, pp. 852-857.
  15. Stove, A.G., 1992.LinearFMCWRadarTechnique,IEE Proceedings F - Radar and Signal Processing, 139(5): 852-857. https://doi.org/10.1049/ip-f-2.1992.0048
  16. Sudhir, K.C., C.S. Yang, K. Ouchi, and P. Shanmugam, 2012. Ship Recognition by Integration of SAR and AIS, The Journal of Navigation, 65(2): 323-337. https://doi.org/10.1017/s0373463311000749
  17. Won, E.S. and K. Ouchi, 2011. Comparison of Ship Detection Algorithms Using ALOS-PALSAR, Ground-Based Maritime Radar and AIS, Proc. Of 20113rd International Asia-Pacific Conference on Synthetic Aperture Radar (APSAR), Seoul, South Korea, Sep. 26-30, pp. 1-4.
  18. Zhao, Z., K. Ji, X. Xing, H. Zou, and S. Zhou, 2014. Ship Surveillance by Integration of Space-borne SAR and AIS - Review of Current Research, The Journal of Navigation, 67(1): 177-189. https://doi.org/10.1017/s0373463313000659